Date: November 16th, Thursday
Time: 11:45 am
Place: Little 339 (the Atrium)
Speaker: Weihong Guo

Drinks and Pizza will be provided after the talk

Title:  Using Non-parametric Distribution Approximation to Smooth and
Segment Images Simultaneously

Abstract: 

Segmentation of anatomical structures from medical images has very important
applications in diagnosis, surgical planning, navigation, and various medical
evaluations.

Medical images are usually corrupted by high level noise, thus, to segment
medical images, smoothing has to be considered. Moreover, image intensity is
usually of complex multi-modal distribution, so to set up model based on a
specific parametric assumption would be erroneous.

Therefore, a non-parametric distribution approximation is applied.
Strength from partial differential equations and statistics are mingled
together to create a fast and efficient framework for simultaneous
segmentation and smooth of medical images. It is able to robustly find
objects whose interiors have high noise level and/or complex multi-modal
intensity distribution. Superiority of the proposed model in segmenting various
images with different levels and types of noise over other models will be
showed.